Elon Musk admits AI is proving very expensive

Elon Musk has admitted that artificial intelligence does not generate enough revenue to cover its enormous energy costs. The move, welcomed by the sector's critics, fuels antiestéticars that the AI bubble could burst.

## Musk's admission

Elon Musk has taken a step many expected and few believed possible: admitting that artificial intelligence is very expensive. In a recent post, the entrepreneur hinted that the sector's revenues do not offset the energy expenditure required to train and maintain the models. It is no minor confession coming from someone who has invested billions in the AI race.

The statement comes at a time when the tech sector is experiencing unprecedented investment fever. Data centres are consuming electricity at a rate that is starting to strain power grids across much of the world. Musk, who until now had championed the technology's potential, is now focusing on the bill.

## An energy bill that doesn't add up

The cost of electricity has become the elephant in the room for AI. Training a large model requires power measured in megawatts, and it is not cheap. Companies in the sector compete to secure supply, and energy prices determine the viability of each project.

The equation is simple to set out and hard to solve: if revenue from subscriptions and licences does not cover the electricity bill, the business is not sustainable. Musk has said so bluntly. And that, coming from someone who has spent years promising that AI will change everything, sounds like a change of script.

## Antiestéticar that the bubble will burst

Warnings about the bubble bursting are nothing new, but they carry more weight when they come from someone on the inside. Some argue the sector is experiencing an overvaluation similar to the dot-com era, with investments that fail to translate into profits. The difference is that this time we are talking about physical infrastructure, cables and watts, not just web pages.

Musk has not set a date for the burst. Others, such as the author of the post that sparked this debate, even suggest it could happen "tomorrow". It is a hypothesis, not a fact. What is a fact is that Musk himself has acknowledged the mismatch between cost and revenue. And in a sector that lives on expectations, that weighs more than any chart.

## Key points

- Elon Musk has admitted that AI does not generate enough revenue to cover its energy costs.
- Data centres' electricity consumption is measured in megawatts and competes for supply.
- Warnings about the bubble bursting are circulating on social media, though with no confirmed date.
- The sector has accumulated billions in investment without clear profits to back it up.